Three indexes of spatial aggregation are developed and used to examine the aggregation pattern of sardine (Sardinops sagax) and anchovy (Engraulis ringens) in the Peruvian Humboldt Current System, determined from 36 acoustic surveys conducted from 1983 through 2003 by the Peruvian Marine Institute (IMARPE). Each index assesses a different aspect of aggregation: the concentration, the percent occupancy of space and the clustering of high-fish abundance. Both time-series correlation and tree-based clustering-regression method, classification and regression trees (CART), were used to relate each of the indexes to environmental variables (season, temperature anomaly and year). Additionally, a measure of onshore-offshore distribution, the average distance from the coast, and abundance variables (the average acoustic backscatter per occupied sampling unit, and the acoustically estimated total abundance of sardine and anchovy from IMARPE) were related to environmental factors by using CART. We show that the 1983-2003 time series can be divided into three different periods: with shifts in 1992 and in 1997-98. Sardine and anchovy showed large differences in both abundance and aggregation among these periods. Although upwelling ecosystems support dramatic and sudden changes in environmental conditions, fish responses are sometimes smoother than usually suggested and there are transition periods with concomitant high biomasses of anchovy and sardine, but with different spatial aggregation patterns. Observed relationships between environmental proxies and aggregation patterns support the habitat-based hypothesis that environmentally mediated alterations in range lead to population changes.
